Data Management for Research

MSU LibrariesResearch Data Management Guidance

Research Data Management

Aaron [email protected]

@aaroncollie

mailto:[email protected]



Introductions

• Please tell us your name and department

• A brief description of your primary research area

• What do you consider to be your research data

• Experience and/or comfort level with managing research data?

cc http://www.flickr.com/photos/quinnanya/


• Introduction• Background

• The Impetus: NSF Data Management Plan Mandate• The Effect: Policy to Practice• The Response: Changing Data Landscape

• Fundamentals Practices• File Organization• Data Documentation• Reliable Backup• Data Publishing, Sharing, & Reuse• Protecting Data & Responsible Reuse

• Data Lifecycle Resources

Agenda


Volunstrordinaries!

Aaron Collie

Hailey Mooney

Devin Higgins

Brandon Locke

Ranti Junus Thomas Padilla

Judy Matthews

Tina Qin


We teach people about RDM

Librarianship

Training

Assessment

Consultation

Ad-hoc

6-12 new clients per semester

100% satisfied / 100% would use again

71% of new clients are referrals

60% requested additional services

15% through NFO, 14% through website


RDM@MSU 101

• Who: You, as the designated steward• What: “the data”• When: Minimum 3 years after

publ./degree• Where: Managed networked storage• Why: Legal, Ethical, Scholarly• How: With fidelity and

documentation sufficient to reproduce the research


http://retractionwatch.com/2014/01/07/doing-the-right-thing-authors-retract-brain-paper-with-systematic-human-error-in-coding/


Jen Doty and Rob O'Reilly, “Learning to Curate @ Emory”. RDAP 2014


Data Management. Isn’t that… trivial?

• Not so much. Data is a primary output of research; it is very expensive to produce high quality data. Data may be collected in nanoseconds, but it takes the expert application of research protocol and design to generate data.

CC-BY-SA-3.0 Rob Lavinsky CC-BY-SA-3.0 Rob


Even more consequential, data is the input of a process that generates higher orders of understanding.

Wisdom

Knowledge

Information

Data

Understanding is hierarchical!

Russell Ackoff


This is the engine of the academic industry…

Defin

e a

ques

tion

Gath

er

info

rmati

on

Form

a

hypo

thes

is

Test

the

hypo

thes

is

Anal

yze

the

data Inte

rpre

t th

e da

ta

Publ

ish

resu

lts

Rete

st


Defin

e a

ques

tion

Gath

er

info

rmati

on

Form

a

hypo

thes

is

Test

the

hypo

thes

is

Anal

yze

the

data

Inte

rpre

t th

e da

ta

Publ

ish

resu

lts

Rete

st


So, things can get a little messy.


Defin

e a

ques

tion

Gath

er

info

rmati

on

Form

a

hypo

thes

is

Test

the

hypo

thes

is

Anal

yze

the

data

Inte

rpre

t th

e da

ta

Publ

ish

resu

lts

Rete

st

The scientific method “is often misrepresented as a fixed sequence of steps,” rather than being seen for what it truly is, “a highly variable and creative process” (AAAS 2000:18).

Gauch, Hugh G. Scientific Method in Practice. New York: Cambridge University Press, 2010. Print. (Emphasis added)


Defin

e a

ques

tion

Gath

er

info

rmati

on

Form

a

hypo

thes

is

Test

the

hypo

thes

is

Anal

yze

the

data Inte

rpre

t th

e da

ta

Publ

ish

resu

lts

Rete

st


The Research Depth Chart

Scientific Method

Research Design

Research Method

Research Tasks

Mor

e Sp

ecifi

c

M

ore

Gen

eric


Defin

e a

ques

tion

Gath

er

info

rmati

on

Form

a

hypo

thes

is

Test

the

hypo

thes

is

Anal

yze

the

data Inte

rpre

t th

e da

ta

Publ

ish

resu

lts

Rete

st

Problem Identification

Study Concept

Literature Review

Environmental Scan

Funding & Proposal

Research Design

Research Methodolog

y

Research Workflow

Hypothesis Formation

Design Validation

Research Activity

Data Management

Data Organization

Data Storage

Data Description

Data Sharing

Scholarly Communication

Report Findings

Publish

Peer Review


Defin

e a

ques

tion

Gath

er

info

rmati

on

Form

a

hypo

thes

is

Test

the

hypo

thes

is

Anal

yze

the

data Inte

rpre

t th

e da

ta

Publ

ish

resu

lts

Rete

st

Problem Identification

Study Concept

Literature Review

Environmental Scan

Funding & Proposal

Research Design

Research Methodolog

y

Research Workflow

Hypothesis Formation

Design Validation

Research Activity

Data Management

Data Organization

Data Storage

Data Description

Data Sharing

Scholarly Communication

Report Findings

Publish

Peer Review






Agenda


Data Management

• The process of planning for and implementing a system of care for your research data before, during, and after a research project in order to ensure a (re)usable resource.


So why are we here?

Good science!Government and Research

Funder Mandates


But why are we really here?

• Impetus: NSF has mandated that all grant applications submitted after January 18th, 2011 must include a supplemental “Data Management Plan”

• Effect: The original NSF mandate has had a domino effect, and many funders now require or state guidelines for data management of grant funded research

• Response: Data management has not traditionally received a full treatment in (many) graduate and doctoral curricula; intervention is necessary


Positive reinforcement….

• National Science Foundation Data Management Plan mandate (January 18, 2011)

• Presidential Memorandum on Managing Government Records (August 24, 2012)–Managing Government Records

Directive: All permanent electronic records in Federal agencies will be managed electronically to the fullest extent possible for eventual transfer and accessioning by NARA in an electronic format.


Positive reinforcement… (cont.)

• White House policy memo (February 22, 2013)– Increasing Access to the Results of Federally Funded

Scientific Research: Federal agencies with more than $100M in R&D expenditures must develop plans to make the published results of federally funded research freely available to the public within one year of publication.

• OSTP policy memo (March 20, 2014)– Improving the Management of and Access to Scientific

Collections: directs each Federal agency that owns, maintains, or otherwise financially supports permanent scientific collections to develop a draft scientific-collections management and access policy within six months.


Positive reinforcement… (cont. w/ teeth!)

• AHRQ = “[unclassified digital data] should be stored and publicly accessible to search, retrieve, and analyze. For sharing of data in digital format, all AHRQ-funded researchers will be required to include a data management plan for sharing final research data in digital format, or state why data sharing is not possible.

• NASA = This plan extends NASA’s culture of open data access to all NASA-funded research.”

• USDA = Phased approach beginning with DMP


Funder Policies

NASA “promotes the full and open sharing of all data”

“requires that data…be submitted to and archived by designated national data centers.”

“expects the timely release and sharing of final research data"

"IMLS encourages sharing of research data."

“…should describe how the project team will manage and disseminate data generated by the project”


Policies for re-use, re-distribution, and creation of derivatives

Plans for archiving data, samples, and other research outcomes, maintaining access

Types of data, samples, physical collections, software generated

• Standards for data and metadata format and content

• Access and sharing policies, with stipulations for privacy, confidentiality, security, intellectual property, or other rights or requirements


• NSF will not evaluate any proposal missing a DMP

• PI may state that project will not generate data

• DMP is reviewed as part of intellectual merit or broader impacts of application, or both

• Costs to implement DMP may be included in proposal’s budget

• May be up to two pages long


• Investigators seeking $500,000 or more in direct costs in any year should include a description of how final research data will be shared, or explain why data sharing is not possible.

• The precise content of the data-sharing plan will vary, depending on the data being collected and how the investigator is planning to share the data.

• More stringent data management and sharing requirements may be required in specific NIH Funding Opportunity Announcements. Principal Investigators must discuss how these requirements will be met in their Data Sharing Plans.


Roles and responsibilities Expected Data Period of data retention• Data formats and dissemination• Data storage and preservation of access


Local Response: Policy

University Research Council Best Practices:Research Data: Management, Control,

and Access– To assure that research data are appropriately

recorded, archived for a reasonable period of time, and available for review under the appropriate circumstances.• Ownership = MSU• “Stewardship” = You• Period of Retention = 3 years• Transfer of Responsibility = Written Request


Broader Response: Changing Data Landscapes

• Data Management Competencies– Standards & Best Practices– Discipline Specific Discourse

• Data sharing and open data– Data sets as publications– Data journals– Citations for data (e.g., used in secondary

analysis)– Data as supplementary materials to traditional

articles– Data repositories and archives


Curation responsibilities (Carlson, The Chronicle, 2006)

“Data from Big Science is … easier to handle, understand and archive.

Small Science is horribly heterogeneous and far more vast. In time Small Science will generate 2-3 times more data than Big Science.”

big science data

small science data

institution?

domain?

MacColl, John (2010). The Role of libraries in data curation. RLG Partnership Annual Meeting, Chicago. June 2010


What’s in it for me?

• Better organization = less headaches– Course management – Bibliographic management– File management– Research

• Career advancement– Publish datasets and list on your CV– Data management is an “unnamed

practice” – name it for yourself and your students!


Data Sharing Impacts

• Reinforces open scientific inquiry

• Encourages diversity of analysis and opinion

• Promotes new research, testing of new or alternative hypotheses and methods of analysis

• Supports studies on data collection methods and measurement

Cc http://www.flickr.com/photos/pinchof_10/


Data Sharing Impacts

• Facilitates education of new researchers

• Enables exploration of topics not envisioned by initial investigators

• Permits creation of new datasets by combining data from multiple sources






Agenda


Research Data Management Fundamentals

• Documentation• File Organization• Storage & Backup• Data Publishing, Sharing,

& Reuse• Protecting Data

& Responsible Reuse


Documentation Practices: Overview

• Researchers benefit from proper documentation to decipher or reuse their datasets – even prior to thinking about sharing

• Think “downstream”


Documentation Practices: Overview1. At minimum create a

README file that you can use to document your project

2. Utilize standards for describing data including Metadata Standards

3. If applicable, use in-line code commentary to explain code

(cc) Will Scullin


Create a README file

• At minimum, store documentation in readme.txt file or equivalent, with data– What data consists of– How it was collected– Restrictions to distribution or use– Other descriptive information


• “Data about data”• Standardized way of describing data • Explains who, what, where, when of

data creation and methods of use• Data more easily found• Data more easily compared to other

data sets

Use Metadata Standards



Basic project metadata:• Title • Language • File Formats

• Creator • Dates • File Structure

• Identifier • Location • Variable List

• Subject • Methodology • Code Lists

• Funders • Data Processing • Versions

• Rights • Sources • Checksums

• Access Information

• List of File Names



• Dublin Core: Commonly-used descriptive metadata format facilitates dataset discovery across the Web.

• Data Documentation Initiative (DDI): Defines metadata content, presentation, transport, and preservation for the social and behavioral sciences.

• ISO 19115:2003: Describes geographic data such as maps and charts.

• More examples:http://www.lib.msu.edu/about/diginfo/collect.jsp

http://www.lib.msu.edu/about/diginfo/collect.jsp

http://www.lib.msu.edu/about/diginfo/collect.jsp


Use In-Line Code Commentary

Example of R code commentary

# Cumulative normal densitypnorm(c(-1.96,0,1.96))

• If applicable, in-line code commentary helps explain code


File Organization Practices: Overview

1. Design a file plan for your research project

2. Use file naming conventions that work for your project

3. Choose file formats to maximize usefulness

“When I was a freshmen I named my assignments Paper Paperr Paperrr Paperrrr”-Undergrad


Design a File Plan

• File structure is the framework• Classification system makes it easier to

locate folders/files• Benefits:– Simple organization intuitive to team

members and colleagues– Reduces duplicate copies in personal

drives and e-mail attachments


Design a File Plan

Choose a sortable directory hierarchy

• Example 1: Investigator, Process, DateCollie

TEI_Encoding20110117

• Example 2: Instrument, Date, SampleUsability Survey

2012043sample_1


Design a File Plan

Example documentation of Directory Hierarchy: /[Project]/[Grant Number]/[Event]/[Investigator/Date]


Use File Naming Conventions

– Enable better access/retrieval of files– Create logical sequences for file sorting– More easily identify what you’re

searching for


• Meaningful but short—255 character limit

• Use alphanumeric characters – Example: abc123

• Capital letters or underscores differentiate between words

• Surname first followed by initials of first name



• Year-month-day format for dates, with or without hyphensExample 1: 2006-03-13Example 2: 20060313

• Decide on a simple versioning methodExample: file_v001



• To create consistent file names, specify a template such as:

[investigator]_[descriptor]_[YYYYMMDD].[ext]


This Not ThissharpeW_krillMicrograph_backscatter3_20110117.tif KrillData2011.tif

This Not ThisborgesJ_collocation_20080414.xml Borges_Textbase.xml


Choose Appropriate File Formats

• Non-proprietary• Open, documented standard• Common usage by research

community• Standard representation (ASCII,

Unicode)• Unencrypted• Uncompressed


Choose Appropriate File Formats

Format Genre Optimal Standards TEXT .txt; .odt; .xml; .html

AUDIO .flac; .wav,

VIDEO .mp2/.mp4; .mkv

IMAGE .tif; .png; .svg; .jpg

DATA .sql; .csv


Storage & Backup Practices

1. Avoid single points of failure

2. Ensure data redundancy & replication

3. Understand common types of storage

(cc) George Ornbo

Data at significant risk of loss without storage and backup plan


Avoid Single Points of Failure

A single point of failure occurs when it would only take one event to destroy all data on a device

• Use managed networked storage when possible

• Move data off of portable media• Never rely on one copy of data• Do not rely on CD or DVD copies to be readable• Be wary of software lifespans


Ensure Data Redundancy

• Effective data storage plan provides for 3 copies:– Primary authoritative copy– Secondary local backup– Tertiary remote backup

• Geographically distribute and secure– Local vs. remote, depending on needed

recovery time • Personal computer, external hard

drives, departmental, or university servers may be used


Ensure Data Redundancy

• Cloud storage – Amazon s3– Google–MS Azure– DuraCloud– Rackspace– Glacier

Note that many enterprise cloud storage services include a charge for in/out of data transfers

$$$


Understand Common Types of Storage

• Optical Media• Portable Flash Media• Commercial Hard Drives• Commercial NAS• Cloud Storage• Enterprise Network Storage• Trusted Archival Storage



• Features of storage types:• Portable data transfers• Short-term storage• Project term storage• Networked data transfer• Long-term storage• Reliable backup option


Understand Common Types of StoragePortable

Data Transfer

Short Term Storage

Project Term Storage

Networked Data Transfer

Long Term Storage

Reliable Backup Option

Optical Media ✔ ✗ ✗ ✗ ✗ ✗

Portable Flash Media

✔ ✔ ✗ ✗ ✗ ✗

Commercial Hard Drives

✔ ✔ ✔ ✗ ✗ ✗

Commercial NAS ✗ ✔ ✔ ✔ ✗ ✗

Cloud Storage ✗ ✔ ✔ ✔ ✗ ✗

Enterprise Network Storage ✗ ✔ ✔ ✔ ✔ ✔

Trusted Archival Storage ✗ ✗ ✗ ✔ ✔ ✔



Media Storage @ MSU

Optical Media MSU Computer Store—Sells Optical Media and hardware accessoriesUAHC Media Storage Service—Offers physical lock-box like storage for MSU

Flash Media MSU Computer Store—Sells Optical Media and hardware accessoriesUAHC Media Storage Service—Offers physical lock-box like storage for MSU

Commercial Hard Drives

MSU Computer Store—Sells Optical Media and hardware accessories.UAHC Media Storage Service—Offers physical lock-box like storage for MSU

Enterprise Cloud Storage

Angel—Free. Ideal for collaboration; not storage space. Phase out 2015Desire2Learn—Free. Ideal for collaboration; not storage space. Replaces AngelGoogleApps—Free. Ideal for collaboration; not intended as storage space

Enterprise Network Storage

AFS Space—Free to 1GB, add’l space can be purchased w/dept. accountIT Services Individual, Mid-Tier and Enterprise Storage—Fee basedHPCC Home or Research—Free up to 1TB. Fee based additions available

Trusted Archival Storage

Disciplinary Repositories – Disciplinary repositories offer archival services for pertinent research data.


Data Publishing, Sharing, Reuse

1. Time-intensive, with potentially high return on investment

2. Publish data in several data publication venues to morebroadly share results of research

Research datasets on par with peer-reviewed journal articles as first-class scholarly contributions


Sharing & Publishing Data• Data preparation for sharing and

publication is a time-intensive process• Potential positive outcomes:• Increased research impact and citations• Enable additional scientific inquiry• Opportunities for co-authorship and

collaboration• Enhance your grant proposal’s

competitiveness


Data Publication Venues• Multiple ways to publish research data• Faculty or project website• Journal supplementary materials• Disciplinary data repository (data

archive)• Varying levels of support for indexing,

access controls, and long-term curation


Data Publication Venues• Disciplinary Data Repository• Securely share data, ensure long-term

access• High visibility• Often offer persistent citations• Availability varies across domains• Databib.org directory


Protecting Data & Responsible Reuse1. Consider how to protect

data and intellectual property rights while encouraging reuse

2. Keep in mind ethical concerns when sharing data

(cc) Will Scullin


Intellectual Property• IP refers to exclusive rights of creators

of works• Individual data cannot be protected by

US copyright• Organization of data such as database,

creative work produced by data, and research instruments used may be protected ©


Intellectual Property• Principal investigator’s institution holds

IP rights• Provide clearly stated license for

producing derivatives, reusing, and redistributing datasets• License under Creative Commons• State if any restrictions or embargos on

use• Provide example of how work should be

cited to encourage proper attribution on reuse

• Document any IP / copyright issues


Ethics & Data Sharing• Keep in mind the following ethical

concerns when sharing your data:• Privacy• Confidentiality• Security and integrity of the data

• For data involving human subjects, obtain written permission or consent stating how the data may be reused


Best Practices = High Impact Data• File organization ensures easier access

and retrieval of data• Documentation makes datasets

accessible and intelligible to users• Storage and backup safeguards data• Data publishing and sharing

encourages the most widespread reuse of data

• Data protection ensures responsible reuse






Agenda


http://www.lib.msu.edu/rdmg




Contact

Aaron [email protected] @aaroncollie



Education

Data Management for Research